Improved control for running advanced agents
An advanced agent runs on all the nodes in a cluster by default. However, if you want to run the agent on only one node, you must manually disable the agent on all the other nodes. Prior to Pega 7.2.2 , if you wanted the agent to not run concurrently on two nodes at a specified time, you had to manually disable the agent on the other node. Now you have additional controls for advanced agents that are configured to run on multiple nodes so that they do not run concurrently at a given time or within a scheduled time.
You can specify the following options for advanced agents on the Schedule tab of the Agents rule form:
- Run this agent on only one node at a time – Select this check box to specify that the advanced agent runs on only one node in a cluster at a specified time.
- Delay the next run of the agent in the cluster for the specified time period – Select this check box to specify the time within which you want the advanced agent to run on only one node in the cluster.
For example, you can specify Block execution within this time windowfor two hours for a cleaner agent that starts running at 10:00 on Node 1. For the next two hours, this advanced agent will not run on any other node in the cluster.
The following table provides additional details about this scenario.
| Advanced agent | Node 1 | Node 2 | Node 3 |
|---|---|---|---|
| Cleaner agent at 10:00 | Started running | Will not run | Will not run |
| Cleaner agent at 10:30 | Completed running | Will not run | Will not run |
| Cleaner agent at 11:00 | Will not run | Will not run | Will not run |
| Cleaner agent at 11:30 | Will not run | Will not run | Will not run |
| Cleaner agent at 12:00 | Can start running | Can start running | Can start running |
- After the specified time elapses, the advanced agent can run on any of the three nodes.
- If the agent on the node in the preceding scenario gives an error, other instances of the agent on any other node in the specified cluster are run on the node's schedule.
- If all nodes in the cluster restart before the next run of an agent, no node in the cluster will run the agent until the block interval between executions has elapsed.